Conference-ALC 07-Oxide Thickness Measurement by Scanning Electron Microscopy with Controlling Ultra-Low Voltage ∗

The secondary electron (SE) images were recorded for Si wafers with and without SiO2 thin (10 nm-50 nm) films by a scanning electron microscope (SEM) with changing primary electron energy from 2.5 keV down to 0.2 keV during one frame scan. Brightness of the SE image increased up to critical values and then decreased with decreasing primary electron energy for the specimens with SiO2 films. The critical voltage is higher for specimen with thicker film. These results are discussed by means of the oxide film thickness, the penetration depth of incident electrons, and the surface potential caused by charging. [DOI: 10.1380/ejssnt.2008.35]


I. INTRODUCTION
Sputter-depth-profiling using surface analysis techniques such as Auger electron microscopy (AES) and Secondary ion mass spectrometry (SIMS) have been practically used for measuring thickness of surface thin layers on a local area.These techniques are time-consuming especially when one obtaines two-dimentinal mapping of the film thickness.
Scanning electron microsopy (SEM) is one of thr most popular techniques to invesitigating sample surfaces in many fields of science and engineering.SEM has been widely used to obtain two-dimentional information of surface, such as topology and the average atomic number.We have investigated material surfaces using ultra-lowvoltage SEM.The distribution of thin oxide layers on steel sheet has been clearly seen as a dark contrast in secondary electron (SE) images using low primery electron energy (E p ) [1,2].The origin of the dark contrast has been explained by the following effect [2]; When the penetration of primary electron beam is limited within the oxide layer, the surface of the oxide layer is charged by the electron beam irradiation.The surface potential caused by the charging has positive sign and is limited only few eV when the E p is less than about 1 keV-2 keV because the secondary electron yield is higher than 1 [3].This charging reduces the emission yield of secondary electrons when kinetic energy of secondary electrons are smaller than the surface potential.This scenario provides us with a possibility of estimating thickness of thin oxide films with low electric conductivity by means of the comparison between the contrasts of SE images recorded with different E p , i.e. different penetration depth of primary electrons.In this paper, we will present preliminary results for SiO 2 films, in which the SE images were recorded with controlling E p including ultra low energy less than 1 keV.

II. EXPERIMENTAL
The SE images were recorded for SiO 2 films with the thicknesses of 10 nm, 19 nm, and 50 nm on Si wafers by a Schottky Emission SEM model LEO 1530 (LEO, now Carl Zeiss, Germany).As a refernece, Si wafer etched by HF was also observed.The E p was changed from 2.5 keV down to 0.2 keV (0.1 keV step) during one frame scan.The primary electron beam was from the direction of the surface normal.Typical beam current is 0.16 nA at E p = 1 keV.Scan area is about 180µm × 135µm and scan speed of primary electron beam was about 0.6 msec/µm.The in-lens type SE detector which has a high sensitivity to the electrons with low kinetic energy [4] was used.The SE images were recorded digitally and average brigthness of SE images was measured for area correspnding to each E p using an image-processing software.
Variable parameters of the objective lense, such as forcus and stigmation, and the detector as well as the position of specimens including the working distance were ajusted at the starting E p .Duraing a frame scan, these parameters were untouched but only the E p was changed.Therefore, it should be noted that the fixed parameters were not optimized for other E p 's except for 2.5 keV.

III. RESULTS
Figure 1 shows SE images for Si-wafer with (a) 50 nm-SiO 2 , (b) 19 nm-SiO 2 , and (c) Si wafer etched by HF, in which E p was changed from 2.5 keV to 0.2 keV with a step of 0.1 keV.The SE images are divided into areas recorded with different E p by the horizontal lines corresponding to time when E p was changed.The top part of each image was recorded with E p of 2.5 keV and the bottom part was recorded with E p of 0.2 keV.For the HF-etched Si wafer, another image were recorded (the insert in Fig. 1(c)) with a reduced gain of image because the brightness of the SE image was saturated at low E p as shown in Fig. 1(c).The  SE image becomes brighter with decreasing E p down to 0.2 keV for the Si wafer etched by HF.For the specimens with SiO 2 films, with the E p decreases, the SE images become brighter at high E p as well as for the HF-etched Si wafer.However images become dark at intermediate E p 's and then become bright again with decreasing E p .
The average brightness of SE image were estimated for each area recorded with each E p by a 8 bits scale.Figure 2 (a) shows the average brightness as a function of E p . Figure 2 (b) shows the normalized data as the brightnesses in the E p rang from 1.5 keV to 2.5 eV becomes the same level.Here, we define that the E p where the brightness of the SE image for the specimens with SiO 2 films starts to dercrease is a critical primery energy (E pc ).The E pc determined as E p where the brigthness starts to deviate from curves for the HF-etched Si wafer are about 1.3 keV, 0.7 keV, and 0.3 keV for SiO 2 films with the thickness of 50 nm, 19 nm, and 10 nm, respectively.It is obiously that the E pc is higher for specimen with thicker SiO 2 film.

IV. DISCUSSION
The dependence of the SE yield on E p for Si was reported [5,6], in which the SE yield increases monotonically with E p decreases down to about 0.3 keV-0.5 keV.No significant changes of the SE yield have been reported at intermediate E p between 0.3 keV and at least 10 keV.The E p dependence of the brightness of SE image for the HF-etched Si (in Fig. 2) is similar to these phenomena of the reported SE yield.We consider that the change of the brightness of SE image relates to the SE yield and can be used as a reference curve of a smooth change of SE yield.
The SE yield of SiO 2 increases monotonically as E p decreases down to about 0.2 keV from 1 keV as well as that for Si [5].The brightness of the SE image for SiO 2 films changes at the E pc higher than 0.5 keV and the E pc changes with thickness of SiO 2 films as shown in Fig. 2.These results indicate that the E pc relates to the thickness of SiO 2 films.The effective penetration range of electrons (R p ), which is the distance over which the electrons travel and deposit their excess energy, was estimated using a empirical relationship, R p ∼ 1000E 1.4 [7].The penetration depth is also estimated from results of a Monte Carlo Simulation.These results are shown in Fig. 3 as a function of E p .The E p 's where these values are close to the thick-  ness of each oxides (0.3 keV-0.5 keV for 10 nm, 0.7 keV-0.8keV for 19 nm, and 1.1 keV-1.5 keV for 50 nm-SiO 2 ) are similar to the E pc ( about 0.3 keV, 0.7 keV, and 1.3 keV, respectively).This result strongly suggests that the E pc corrsponds to the E p which the penetration of the incident electrons reachs to the interface between the SiO 2 films and Si substrates.The total SE yield for SiO 2 films are larger than 1 at the E p range lower than at least 3 keV [8].
For the E p lower than E pc , the SE yield is reduced by the positive surface potential caused by the charging because the penetration of primary electrons is limitted within the insulating SiO 2 layers.Therefore, as mentioned in the introduction, the SE image becomes darker due to the positive charging in this situation [2].When the E p exceeds the E pc , the primary electrons reach to the Si substrate.The surface potential becomes equal or near to that of the substrate and therefore hardly restricts the SE emission.
Our results demonstrate the possibility that one can estimate the thickness of SiO 2 film using SEM by evalu-ating the E pc and the effective penetration range.This technique can be extended to the surface layers composed by the materials with low electric conductivity.The E pc is determined by monitoring the change of the brightness of SE with changing E p as shown in Fig. 2.However, the changes of brightness around the E pc are always not sharp but are broad transition in our results.This may be due to many contributions; the distribution and density of primary electorons in the films, the transport properties of electrons between the surface and the substrate.The broad transition makes the determination of the E pc difficult.The investigations of theoretical and experimental approachs are highly required for quantitative estimation of thickness of thin surface layer using SEM.From experimental point of view, the optimaization of SEM parameters is needed to controlling surface charging.It is also expected that the system that allow direct SE intensity measurement will be developed in the future.

V. CONCLUDING REMARKS
We proposed a method to estimate thickness of thin surface layers with low electric conductivity using SEM and demonstrated the possibility by presenting the results for thin SiO 2 films.This technique uses the penetration depth of the primary electrons as shallow as the thickness of surface layer (less than few tens nanometers).It should be emphasized that the low E p is one of key points because the typical E pc is 0.3-1.3keV for SiO 2 films with the thickness of 10 nm-50 nm.

FIG. 2 :
FIG. 2: (a) Average brightness of SE images as a function of Ep.(b) Normalized average brightness as the values becomes almost same in the Ep range from 1.5 to 2.5 eV.The roughly estimated critical values (Epc) are denoted by arrows for the specimens with SiO2 films.

FIG. 3 :
FIG. 3: Attenuation range of electrons as a function of Ep.Estimation of the values is mentioned in the text.